[J3] Vector subscripts and INTENT(INOUT)

Malcolm Cohen malcolm at nag-j.co.jp
Thu May 24 04:00:29 EDT 2018

> naturally interpreted as a technical change.


That would be wrong, this is the whole reason we have an Introduction that lists the technical changes, so that people don’t have to grovel through the implications of wording changes in 600 pages of standard to have any chance of working out if something actually changed.


Even without editorial changes to confuse things, unannounced changes have a long history of not being implemented for quite a few years because no-one realised there was a change!


We have been operating under the scheme: “if the Introduction does not mention a change, and there appears to be a technical change, the standard is Defective and interpretation processing is needed to resolve the issue” ever since the publication of Fortran 2008.  That scheme remains – if the Intro does not mention a change, either the Intro is wrong or the change is wrong or illusory.


>In fact, there is a minor technical change in this case,


No there is not.


> in that the old sentence did not account for the case of the dummy argument having the VALUE attribute.


Yes it did.  Really!  In the case of the VALUE attribute, the effective argument is not the actual argument, and so cannot have a vector subscript.


(It’s true that Fortran 2003 did not have the “effective argument” wording, but Fortran 2003 did not have VALUE arrays either so the issue never arose.)


> Additional limitations related to the use of an array section with a vector subscript as an actual argument


I do not see a good reason for pointing this out when vector-subscripted array sections have *many* limitations.  Pointer association, argument association, storage association, construct association… the fact is that you either can’t use them, or when you do they become read-only, in nearly every form of association we have.


So this would end up as something like “Vector-subscripted array sections have limitations on their uses in different contexts; such limitations are described in those contexts.”… ok that could be better phrased, but anyway it is (or should be) so unsurprising as to be a NSS moment.


IMO it is really not worth the time and energy.  I have yet to see, or think of myself, a form of words that even rises to the level of acceptable.  I further think we have lots of actual issues to discuss, so I would put this at a very low priority at most.




..............Malcolm Cohen, NAG Oxford/Tokyo.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.j3-fortran.org/pipermail/j3/attachments/20180524/7e467aea/attachment.html>

More information about the J3 mailing list